【作者】 柳林霞； 陈杰； 陈文颉；

【Author】 LIU Lin-xia, CHEN Jie, CHEN Wen-jie (Beijing Institute of Technology , Department of Automatic Control, 10081)

【机构】 北京理工大学自动控制系；

【摘要】 针对模板匹配方法,当待识别的物体发生旋转、比例、平移的变换时它不能正确地识别目标的缺点,本文介绍了几种不随物体位置与姿态变化而变化的特征量,主要有不变矩、傅立叶描述子及神经网络等,并比较了这几种特征抽取方法的优劣。更多还原

【Abstract】 Aimed at the difficulties of the template matching recognition method which brought out by the rotation, motion and scale changing of object. this paper introduces several kind features not changing as the pose of object changes. There are invariant momentum. Fourier descriptors and ANN, included in this paper. These methods are also compared in this paper.更多还原